Special Report
What’s Real About
Virtual Reality?
A
s usual with infant technologies, realizing the early dreams for virtual reality
(VR) and harnessing it to real work has taken longer
than the initial wild hype predicted. Now, finally, it’s
happening.
In his great invited lecture in 1965, “The Ultimate Display,” Ivan Sutherland laid out a vision1 (see the sidebar), which I paraphrase:
Don’t think of that thing as a screen, think of it as
a window, a window through which one looks
into a virtual world. The challenge to computer
graphics is to make that virtual world look real,
sound real, move and respond to interaction in
real time, and even feel real.
This personal assessment of
the state of the VR art
concludes that whereas VR
almost worked in 1994, it
now really works. Real users
This research program has driven
the field ever since.
What is VR? For better or worse,
the label virtual reality stuck to this
particular branch of computer
graphics. I define a virtual reality
experience as any in which the user is
effectively immersed in a responsive
virtual world. This implies user
dynamic control of viewpoint.
Frederick P. Brooks, Jr.
University of North Carolina at Chapel Hill
Europe. Every one of the component technologies has
made big strides. Moreover, I found that there now exist
some VR applications routinely operated for the results
they produce. As best I can determine, there were more
than 10 and fewer than 100 such installations as of
March 1999; this count again excludes vehicle simulators and entertainment applications. I think our technology has crossed over the pass—VR that used to
almost work now barely works. VR is now really real.
Why the exclusions? In the technology comparison between 1994 and 1999, I exclude vehicle simulators and entertainment VR applications for different
reasons.
Vehicle simulators were developed much earlier and
independently of the VR vision. Although they today
provide the best VR experiences available, that excellence did not arise from the development of VR technologies nor does it represent the state of VR in general,
because of specialized properties of the application.
Entertainment I exclude for two other reasons. First,
in entertainment the VR experience itself is the result
sought rather than the insight or fruit resulting from
the experience. Second, because entertainment
exploits Coleridge’s “willing suspension of disbelief,”3
the fidelity demands are much lower than in other VR
applications.
routinely use production
applications.
VR almost worked in 1994.
In 1994, I surveyed the field of VR in
a lecture that asked, “Is There Any
Real Virtue in Virtual Reality?”2 My assessment then
was that VR almost worked—that our discipline stood
on Mount Pisgah looking into the Promised Land, but
that we were not yet there. There were lots of demos and
pilot systems, but except for vehicle simulators and
entertainment applications, VR was not yet in production use doing real work.
Net assessment—VR now barely works. This
year I was invited to do an up-to-date assessment of VR,
with funding to visit major centers in North America and
16
Technologies
Four technologies are crucial for VR:4,5
■ the visual (and aural and haptic) displays that
immerse the user in the virtual world and that block
out contradictory sensory impressions from the real
world;
■ the graphics rendering system that generates, at 20
to 30 frames per second, the ever-changing images;
■ the tracking system that continually reports the position and orientation of the user’s head and limbs; and
■ the database construction and maintenance system
for building and maintaining detailed and realistic
models of the virtual world.
November/December 1999
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
0272-1716/99/$10.00 © 1999 IEEE
Four auxiliary technologies are important, but not
nearly so crucial:
■ synthesized sound, displayed to the ears, including
directional sound and simulated sound fields;
■ display of synthesized forces and other haptic sensations to the kinesthetic senses;
■ devices, such as tracked gloves with pushbuttons, by
which the user specifies interactions with virtual
objects; and
■ interaction techniques that substitute for the real
interactions possible with the physical world.
Sutherland’s 1965 Vision
Display as a window into a virtual world
Improve image generation until the picture looks real
Computer maintains world model in real time
User directly manipulates virtual objects
Manipulated objects move realistically
Immersion in virtual world via head-mounted display
Virtual world also sounds real, feels real
Table 1 summarizes the technology progress since
1994.
tion, panoramic projection, and workbenches. Each has
its own peculiar merits and disadvantages.
Displays
Head-mounted displays. The most noticeable
advances in HMDs have occurred in resolution, although
color saturation, brightness, and ergonomics have also
improved considerably. In 1994, one had a choice of costly and cumbersome CRT HMDs, which had excellent resolution and color, or economical LCDs, which had coarse
resolution and poor saturation. Today economical LCDs
have acceptable resolution (640 × 480 tricolor pixels)
and good color saturation.
Field of view still poses a major problem with HMDs,
with 45-degree full-overlap about the industry median,
at prices in the $5,000 range.
Display technology has advanced very rapidly, pulled
along by the television, presentation-projection, and
LCD-device markets, rather than just the still-small VR
market. As VR was developing, much ink was spilled
over the relative merits of various formats of displays:
head-mounted displays (HMDs), CAVE-like (Cave Automatic Virtual Environment) surround projectors,
panoramic projectors, workbench projectors, and desktop displays.
Most workers consider desktop displays not to be VR
because they
■ hardly block out the real world,
■ do not present virtual-world objects in life size, and
therefore
■ do not create the illusion of immersion.
To be sure, one could say the same for workbenches,
but somehow the result is not at all the same. Workbenches were originally designed for human-body models, which they do display life-size. Moreover, they
subtend a large visual angle and hence achieve substantial real-world blocking. When used for battle planning, for example, the workbench in fact represents in
life size the sandtable devices they displace.
The debate about display format seems to be over. People recognize that each format proves clearly superior
for some set of real applications, yet none dominates. In
our laboratory at the University of North Carolina at
Chapel Hill, we use them all—HMDs, surround projec-
CAVEs and kin. Many major VR installations now
use the surround-projection technology first introduced
in the University of Illinois-Chicago Circle CAVE. From
three to six faces of a rectangular solid are fitted with
rear-projection screens, each driven by one of a set of
coordinated image-generation systems.
Standard off-the-shelf projector resolution is now up
to 1280 × 1024, better than SVGA (super video graphics
array) and even XGA (extended graphics array). In a 10foot cave, this gives an angular resolution of about 4
minutes of arc; human visual acuity is 1 minute of arc
for 20/20 vision.
The principal advantages of surround-projection displays are
■ a wide, surrounding field of view, and
■ the ability to give a shared experience to a small group
(of whom one or none are head-tracked).
Table 1. Progress in VR technologies.
1994
1999
Swimming due to lags
Displays: narrow field of view
or poor resolution or high cost
Limited model complexity
Poor registration in augmented reality
Still a major problem for HMDs and caves, not screens
HMD resolution of 460 by 680 real pixels now affordable
Projector resolution 1280 by 1024
Rendering limited mostly by cost
Augmented reality dynamic registration still hard;
1 ms = 1 mm error
Wide-area tracking available; wireless not yet in use
Ergonomics getting there
Model engineering a major task
Tethered ranging
Bad ergonomics
Tedious model building
IEEE Computer Graphics and Applications
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
17
Special Report
The principal disadvantages are
■ the cost of multiple image-generation systems,
■ space requirements for rear projection,
■ brightness limitations due to large screen size, which
results in scenes of approximately full-moon brightness and hinders color perception,
■ corner and edge effects that intrude on displayed
scenes, and
■ reduced contrast and color saturation due to light
scattering, especially from opposing screens.
This latter problem, which I notice in UNC’s surroundprojection VR facility and in all I have visited, seems
inherent in the geometry itself. However, careful choice
of screen material (Carolina Cruz-Neira of Iowa State
University has an unpublished study on this) and probably polarizing glasses can ameliorate the problem. The
brightness, contrast, and color saturation problems are
serious enough that the team at the Fraunhofer Institute at Stuttgart reports that their client automobile stylists and industrial designers have rejected their cave in
favor of Fraunhofer’s panoramic display installation.
The users do insist on a life-sized display.
Panoramic displays. One or more screens are
alternatively arranged in a panoramic configuration.
This suits groups especially, and multidisciplinary
design reviews commonly use this type of display. One
person drives the viewpoint.
Workbenches. The workbench configuration lays
a rear-projection screen flat and positions the projector
so that the workbench’s length approximates that of a
human body. One, two, or conceivably more tracked
viewers each perceive a custom-generated image.
Angular resolution typically is about 4 minutes of arc
near the center of the display. Since the eye-to-farscreen-border plane limits the apparent height of an
object, many workbenches can be tilted to resemble
drafting tables.
Rendering engines
Rendering engines have benefited from significant
advances in speed and reductions in cost.
Speed. Rendering engines have improved radically
in the past five years (almost four of Gordon Moore’s 18month performance-doubling periods). The complexity of the virtual worlds that could be visualized was
sharply limited in 1994, when the fastest commercially
available engines had actual speeds of about 600 K polygons per second, or about 30 K polygons in a 1/20second frame. Today each pipe of an 8-pipe,
32-processor SGI Reality Monster can render scenes of
up to 180 K polygons in 1/20 second. Moreover, much
larger configurations are available.
In one sense, world-model complexity is more dollarlimited than technology-limited today. In practice, world
models containing more than 500 K polygons still
require algorithmic attacks to achieve real-time
rendering.
18
Cost. Meanwhile, steady progress in mass-market
CPUs yields multi-hundred-megahertz clock speeds.
The progress of graphics accelerator cards, driven by
the game market, has matched this CPU progress. Consequently, VR configurations of quite acceptable performance can now be assembled from mass-market
image-generation engines. For many applications,
image generation no longer dominates system cost.
Tracking
In 1994, tracking the viewer’s head motion was a
major problem. Tracker ranges tethered the viewer to
an effective radius of about four feet. Tracker accuracy
suffered from severe field distortion caused by metal
objects and magnetic fields.
Unlike display technology and image-generation technology, tracking technology has not had a substantial
non-VR market to pull it along. The most important collateral market has been motion capture for entertainment applications, and that market has not pressed the
technology on accuracy. So progress in tracking has not
matched that of displays and image generation.
Nevertheless, progress has occurred. The UNCChapel Hill outward-looking optical tracker is routinely operated over an 18 × 32-foot range, with accuracy of
1 mm and 0.1 degree at 1,500 updates per second. Commercial off-the-shelf (COTS) trackers give working
range radii of 8 to 10 feet. Hybrid-technology trackers
seem most promising, combining inertial, optical, ultrasonic, and/or magnetic technologies.
Ergonomics
With wide-range trackers, another problem
emerges—the wires to the user. The physical tether, not
too much of a bother when the user was electronically
tethered anyway, becomes an ergonomic nuisance once
the user can walk around in a room-sized area.
In principle, substituting wireless links for wires can
solve this problem. For users in surround-projection systems, COTS wireless systems serve, since only tracking
and button-push data need to be transmitted.
For users with HMDs, the problem becomes much
more serious—two channels of high-definition video
must also be transmitted, and COTS wireless systems
do not yet have the required portability. HMDs also
require body-mounted power for free-ranging viewers to
wear them.
System latency
Perceptually, the greatest illusion breaker in 1994 systems was the latency between user motion and its representation to the visual system. Latencies routinely ran
250 to 500 ms. Flight simulator experience had shown
latencies of greater than 50 ms to be perceptible. In my
opinion, end-to-end system latency is still the most serious technical shortcoming of today’s VR systems.
In HMD systems, head rotation is the most demanding motion, with typical angular velocities of 50
degrees per second. Latencies of 150 to 500 ms make
the presented scene “swim” for the user, seriously damaging the illusion of presence. In projection systems,
whether surround, panoramic, or workbench, the
November/December 1999
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
viewer’s head rotation does not change the generated
image—only viewpoint translation, and any motions
of the interaction device and of virtual objects. Hence,
system latencies are not so noticeable in these systems—but they nevertheless damage the illusion of
presence. I see latencies of 150 to 250 ms being accepted without outcry.
Many advances in image rendering speed came
through graphics processor pipelining, which has hurt
system latency. The designers of tracking systems have
given insufficient attention to system latency. Moreover, many VR systems have been pieced together using
standard networking between the tracker’s computer
and the image generator, and this contributes noticeably to latency.
The latency problem becomes extremely serious in
augmented reality systems in which the virtual world
superimposes on the real world. The challenge lies in
dynamic registration—the two worlds should stay rigidly locked together both for the illusion of reality and for
task performance. Holloway has studied the viewer
motions of a cranio-facial surgeon while examining a
patient and developing an operating plan. For those
viewer motions, Holloway found that a millisecond of
latency translated into a maximum registration error of
one millimeter in, for example, superimposing CT scan
and/or MRI scan data on the patient’s visible face, as
perceived by the surgeon through an HMD.6 Since the
application required millimeter accuracy, and today’s
best systems have achieved latencies of at best 40 to 50
ms, Holloway chose to pursue the cranio-facial application no further.
The most exciting, although not the easiest, augmented reality applications are surgical. One attack uses
video camera technology to acquire the real-world
image, which is then video-combined with the virtual
image. This approach has in principle two advantages
over optical combination—the video real-world image
can be delayed to match the virtual image, and obscuration of far objects by near ones can be done symmetrically between the images.
In my opinion, end-to-end system latency
is still the most serious technical
shortcoming of today’s VR systems.
a substitute, such as the spice rack and contents, or stove
knobs.)
I have found a breadth-first iterative-refinement strategy best for modeling. First, make a simple representation of each major object, then of each minor object.
Then do a level of refinement on each, guided by the
eye—What approximation hurts worst as one experiences the world?
Textures are extremely powerful, as SGI’s Performer
Town first demonstrated. Image textures on block models yield a pretty good model rather quickly. Moving
through even a rough model wonderfully boosts the
modeler’s morale and enthusiasm, as well as guiding
refinement.
Buying models. Several firms offer catalogs of
models, at different levels of detail, of common objects
including people. It is cheaper, and faster, to buy than
to build these components of a custom world model.
Inheriting CAD models. By far the simplest way
to get superbly detailed models of designed objects,
whether existing or planned, is to get the computerassisted design (CAD) data. But even that rarely proves
simple:
■ We have found it very difficult to get the original com-
Model engineering
Now that we can explore quite large virtual world
models in real time, we find that acquiring, cleaning,
updating, and versioning even static world models is
itself a substantial engineering task. It resembles software engineering in magnitude and in some, but not all,
other aspects.My own rule of thumb is that managing a
model of n polygons is roughly equivalent to managing
a software construct of n source lines.
■
■
■
Model acquisition. VR practitioners acquire models in one of three ways: build them, inherit them as
byproducts of computer-aided design efforts, or acquire
them directly by sensing existing objects.
In spite of a variety of excellent COTS tools for model
building, it is tedious and costly work, especially when
accuracy is important. We have found it takes several
man-years to make a model of an existing kitchen that
aims at quarter-inch accuracy. (We do not actually aim
for that resolution where a textured image will serve as
■
■
putational-solid-geometry data, which best encapsulates the designer’s thoughts about form, function,
and expected method of construction. We usually
receive an already tessellated polygonal representation. Quite often, it will be severely over-tessellated,
needing polygonal simplification to produce alternative models at different levels of detail.
Formats will almost always have to be translated. Do
it in an information-preserving way.
CAD models often require substantial amounts of
manual cleaning. In some CAD systems, deleted,
moved, or inactive objects stay in the database as
polygonal ghosts. Coincident (twinkling) polygons
and cracks are common.
AutoCAD does not capture the orientation of polygons.
Orienting them takes automatic and manual work.
If any of the subsequent processing programs requires
manifolds, making them from CAD data will take
much work.
CAD models, particularly of architectural structures,
typically show the object as designed, rather than as
built.
Models from images. For existing objects only, as
opposed to imagined or designed objects, imaging can
IEEE Computer Graphics and Applications
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
19
Special Report
yield models. Imaging may be done by visible light, laser
ranging, CAT and MRI scans, ultrasound, and so forth.
Sometimes one must combine different imaging modalities and then register them to yield both 3D geometry
and visual attributes such as color and surface textures.
Recovering models from images is a whole separate
technology and an active research area, which I cannot
treat here.
■
■
■
■
Vehicle simulation—first and still much the best
Entertainment—virtual sets, virtual rides
Vehicle design—ergonomics, styling, engineering
Architectural design and spatial arrangement; submarines, deep-sea oil platforms, process plants
■ Training—only the National Aeronautics and Space
Administration
■ Medicine—psychiatric treatment
■ Probe microscopy
Applications
The most important thing in VR since 1994 is not the
advances in technologies, but the increasing adoption
of its technologies and techniques to increase productivity, improve team communication, and reduce costs.
VR is now really real.
Finding the production-stage applications
For my assessment, I broadcast e-mail messages to
the mailing list for the Virtual Reality 99 Conference and
to the United Kingdom VR Special Interest Group (ukvrsig), inviting people to send reports of VR applications.
I defined three stages of application maturity:
■ Demonstration
■ Pilot production—has real users but remains in the
developers’ hands, under test
■ Production—has real users doing real work, with the
Of these, vehicle simulation was the first application of
what today we call VR. It is not only the oldest, it is also
the most advanced. The results accomplished in vehicle
simulation provide both an existence theorem and a
challenge for other applications: It is possible to do VR
exceptionally well, and it pays. Surprisingly, there has
been relatively little knowledge interchange between
the vehicle simulator discipline and the new VR discipline, due I think to ignorance and sloth on the part of
the VR community.
A second batch of for-profit production applications of
VR lie in the entertainment sphere. By and large, the most
elaborate use of computer graphics in entertainment has
been non-real-time graphics, used for animation and special effects, rather than VR. However, theme parks and
arcades are increasingly installing true VR experiences. I
shall not treat these applications at all here.
system in the users’ hands
Flying a 747 simulator at British Airways
I especially invited reports of applications in the pilot
and production stages.
In the discussion here, I report on some productionstage applications, that is, employed in dead earnest by
users—not developers—for the results gained, rather
than as an experiment or trial. This report does not aim
to describe all such applications. On the other hand,
most published reports do not sharply distinguish the
stages of progress of systems. This report describes
applications that I know to be in production. In most
cases I visited the application site, verified the application’s status, and tried to learn what unexpected experiences and lessons have resulted from the process of
making them work. In a sense, this spells out “what’s
real” about some VR applications.
The most surprising result of my call for information
on VR projects is how few I could find in mid-1998 that
were really in routine production use. Some, of course,
are not discussed openly for competitive or security reasons. A great many more were in pilot status, well past
the demo stage and almost to the production stage. I
believe this situation to be changing very rapidly and
that another five years will find VR applications such as
those I describe to be much more widespread.
For example, the use of immersive VR for visualization of seismological data has been intensively pursued
both in oil companies and in universities. One company, Landmark, is even making major VR installations in
the US and in Europe for that purpose. Yet, as far as I
could learn by inquiries, no one was quite yet using VR
in routine seismic interpretation.
Here are the kinds of verifiable production applications I found:
20
To my delight, I got to spend an hour flying a 747 simulator at British Airways’ facility at Heathrow. It was a
stunningly good illusion—the best VR I have ever experienced. Rediffusion Ltd. built the simulator, which cost
about $13 million.
The visuals appear on a spherical zonal screen about
12 feet from the pilot and copilot seats. The physical
setup faithfully models the interior of the cockpit, and
the instruments faithfully display the results of real
avionics—the steering yoke seems to fight back properly. The whole pilots’ cabin, plus room for the instructor,
is mounted on a motion platform that gyrates within a
three-story space. The vehicle dynamics and simulated
motions appear to be of very high quality. The sound is
superb: engine sound, wind sound, taxiing bumps in the
pavement, radios.
Within a very few minutes I was not in a simulator, I
was flying the airplane: taxiing, taking off, climbing out,
circling the airport, and trying to keep the plane at constant altitude. (The 747’s dynamics differ markedly from
those of the light planes in which I learned to fly years
ago.) So compelling was the illusion that the breaks in
presence came as visceral, not intellectual, shocks. One
occurred when the instructor abruptly changed the
scene, from circling above London to an approach to
Hong Kong. The other occurred when I taxied up to a
hangar in Beijing and looked back (to about 4 o’clock)
to ensure that I would not brush wingtips with a parked
aircraft—and the view was just empty gray! The projected visuals did not reach that far around.
Lessons. My host, Michael Burtenshaw, explained
that their successes drove the steady evolution and uni-
November/December 1999
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
Courtesy Warsash Maritime Centre, Southampton Institute
versal adoption of flight simulators
to train pilots on new aircraft types.
At Heathrow, British Airways now
has 18 simulators in four buildings,
each specialized to an aircraft type.
These they keep busy training both
their own pilots and pilots of smaller airlines, to whom they sell instruction. Simulators, though costly,
are much cheaper than airplanes.
Much more important, pilots can
train and exercise in extreme situations and emergency procedures
where real practice would imperil
aircraft and lives. Many major airlines have similar sets of simulators;
so do various air forces.
Increasingly, real avionics, which
make up a good chunk of the cost of
high-end simulators, are being
replaced by individual PCs. Each PC
simulates one instrument’s behavior and drives its output dials. Significant economies
result.
British Airways needs to keep a simulator type as long
as it keeps the corresponding aircraft type—up to 25
years. I found it memorable to walk through the facility and see old computers running whose existence I had
almost forgotten and whose maintenance is a nightmare for the airline: Xerox Sigma 3, DEC PDP-11, and
Vax 11-780.
Merchant ship simulation at Warsash
The Warsash Maritime Centre of Southampton Institute trains both deck and engineering officers for the
merchant marine. It offers a two-year cadet course at
the technical-school level. A principal undertaking,
however, is a set of one-week short courses designed for
both the qualification testing and skill enhancement of
deck and engineering officers. Shipping companies regularly send deck or engineering teams of four officers to
train together, to build team skills. The facility operates
routinely as a revenue-producing element of Southampton Institute. David Gatfield served as my host.
Warsash runs a rich set of simulators: an engine-room
control system, a liquid-gas cargo-handling simulator,
a first-class single-bridge simulator, and a coupled simulator consisting of three bridge simulators, each capable of handling a full deck-officer team, for multiship
maneuvers. The best bridge simulator represents a
generic merchant vessel, although the view of the ship
from the bridge can be customized (via texture mapping) to represent a particular ship type or even a particular vessel. Similarly, the look and feel of the controls
can be customized to a small degree, to simulate different bridge configurations (see Figure 1).
The visual surround, approximately 180 degrees, is
generated by seven behind-screen Barco projectors. The
imagery is generated with 768 × 576-pixel resolution
by a set of PCs with accelerator cards. Woofers mounted under the floor do a quite convincing job of simulating engine noises for several different power-plant
1 Ship bridge
simulator at
Warsash Maritime Centre.
types; other speakers provide wind, wave, buoy, and
ship-whistle noises.
The ocean simulation provides a variety of sea states;
the sea model includes tides and currents. Atmospherics, including a variety of visibility and fog conditions,
are effective. An auxiliary monitor provides the function of binoculars—a higher-resolution, restricted fieldof-view image—without the aiming naturalness of true
binoculars. Radar, Loran, geographic positioning system (GPS), depth (fathometer), over-the-ground speed
indicator, and other instruments are faithfully simulated. Norcontrol built the simulator, which cost approximately £2 million in 1995.
I experienced a ferryboat trip from a southern Norwegian port, at twilight. As wind speed and sea state
rose, the most surprising effect was the realism of the
vessel’s pitch and roll, achieved entirely by manipulation of the imagery. My colleagues and I found ourselves
rocking to and fro, and side-to-side, to compensate for
the ship motion. One recent visitor, a professional in ship
simulation, asked the Warsash people if they had much
trouble with their hydraulics. He was surprised when
told there were none.
A fascinating non-VR team-training simulator at
Warsash consists of a small fleet of 1/12 to 1/25 scalemodel merchant ships navigated around a 13-acre lake
with some 30 ship-handling hazards or stations. Each
ship seats two persons, canoe-style. The master’s eyes
are exactly at the height of the bridge. He gives oral commands to the pilot, who actually handles the controls.
The ship scaling results in a seven-fold scaling-up of the
natural breezes and winds.
Wish list. First and foremost, our hosts want morenatural binoculars—an important item often mentioned
in student debriefings. Today’s commercial off-the-shelf
technology offers such capability—it just takes money.
Second, they want better screen resolution, and third,
more screen brightness. Indeed, the simulator worked
well for twilight and night scenes, but could not begin to
IEEE Computer Graphics and Applications
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
21
Special Report
Courtesy of Daimler-Chrysler
2 View of
virtual windshield wiper
visibility at
DaimlerChrysler’s Technology Center.
approximate either the brightness or the dynamic range
of a sunlight-illuminated scene.
Lessons. As with flight simulators, our hosts report
several advantages of simulation over real-ship training:
■ Emergency scenarios, even extreme ones, can be thor-
oughly exercised.
■ Scenarios can readily be run, accelerated, and
switched, enabling more significant experience time
per hour of training.
■ Cost.
A few engineering officer teams report that Warsash’s
large control-board simulator seems dated—their own
engine rooms now have glass-cockpit-type controls,
everything displayed and actuated via computer
screens. Warsash is currently updating the engine-room
simulator to reflect this type of control system. I wonder at the human factors effects of seeing everything at
once via a visual scan versus having to act to bring up
information. Older may be better.
I am convinced that much of the sense of presence
and participation in vehicle simulators comes from the
fact that one can reach out and touch on the simulator
everything reachable on the real vehicle—the near-field
haptics are exactly right.
The second take-home lesson for me from experiencing these working simulators is the extreme importance of getting sound right.
Ergonomics and design at Daimler-Chrysler
For years, VR researchers have worked toward making VR an effective tool for product design and design
review. Today the automobile industry seems way ahead
in adopting the technology for design applications—
most major automobile manufacturers have installations or use nearby ones. Some are in routine production
status.
I visited one such production system, Daimler-
22
Chrysler’s installation in the Small
Car Platform Advance Vehicle Engineering area of their Auburn Hills,
Michigan, Technical Center. Ken
Socks, Don Misson, and Josh Davidson served as my hosts. The configuration includes a high-resolution
stereoscopic Boom by FakeSpace
Systems, worn on the user’s head as
a head-mounted display. The Boom
mechanism provides high-accuracy,
extremely low-latency mechanical
tracking of the user’s head pose. The
user sits in a “buck”—a real car seat,
complete with all its adjustments,
combined with a real steering wheel
and a mocked-up instrument control panel.
Imagery comes from an SGI Onyx
Infinite Reality system. It drives not
only the Boom but also a monocular
projector that displays on a 4-foot by
6-foot screen at one end of a nearby conference table
that seats eight. An Ascension magnetic tracker tracks
the positions of the driver’s hands (but not the fingers)
and of auxiliary objects such as a coffee cup. A short calibration sequence establishes the length and position of
the extended index finger relative to the hand tracker.
Modal buttons enable the driver to cycle among the conformations of the hand avatars—reaching, holding an
object, grasping the steering wheel, and so on.
This installation sees routine weekly use for production design work. I talked with user engineers from
groups doing ergonomic studies (driver controls, access
from front passenger seat, view, and so forth), windshield wiper design (visibility as a function of body size;
see Figure 2), interior trim design (color studies, among
others), and painting studies (detecting accidental
appearances of exterior paint in the interior, due to
metal wrapping). The windshield wiper engineers, for
example, report that the system obviates many arguments—a whole team sees together the visibility as perceived by drivers of different body sizes and shapes.
Lessons. A major factor in the success of the Daimler-Chrysler installation: it was desired, funded, specified, and is operated by a user group who really wanted
it, the Ergonomics activity within Small Car Platform
Advance Vehicle Engineering. They enlisted the (essential) help of the Technical Computer Center in realizing
their system. It offers a prime example of user-pull versus technology-push. Consequently, system fixes and
enhancements follow the users’ priority order, and the
corporate know-how for using the system most effectively develops within one of the user groups. Interestingly, use now extends to many departments other than
Ergonomics, and the system is used as much for other
aspects of design as for ergonomics.
Models provide a second important lesson from the
Daimler-Chrysler experience. Some years ago, their laboratories operated a variety of CAD systems. The Engineering Vice President ordered that all CAD would move
November/December 1999
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
to one system; they selected Catia from Dassault Systemes. Since Catia is the design medium, the Catia models of automobiles under design are current, accurate,
and maintained—the metal is cut from those models.
This means that one of the hardest problems in VR—
how to get accurate, detailed models—does not arise
for the Daimler-Chrysler applications. The models are
byproducts of the design process, and they are furnished
by someone other than the VR systems team.
Experienced engineering organizations have quite formal change-control processes.
The design and design review processes as described
by my hosts at Electric Boat corresponded closely to
those described to me at Newport News Shipbuilding in
Newport News, Virginia, who make aircraft carriers and
other ships, and at Brown and Root Engineering Division of Halliburton in Leatherhead, Surrey, UK, who
design off-shore oil platforms.
Design review at Electric Boat
Lessons. All the engineering organizations perceive
large advantages from VR walkthroughs as a part of
design reviews.
In one review at Electric Boat, a capital tooling person from the factory pointed out that a certain massive
semicylindrical tank could be better fabricated as a cylinder—for which they had extant tooling—then cut in half
and roofed, saving thousands over the engineer’s proposed fabrication from piece parts.
In a multidisciplinary design review at Brown and
Root, the painters from the maintenance force remarked
that certain fixtures on an oil platform should be made
out of extra-heavy-gauge steel because “We can paint
its interior once in the shop, but we’ll never again be able
to paint it after you’ve installed it like that.” The designers chose to change the configuration.
In describing their two years of experience, my hosts
at Brown and Root—Arthur Barker and Martin
Williams—said that the most important effects of their
installing and using their VR theater and its associated
telecommunications was not effects on designs, but
effects on the design processes, in particular the communication of ideas between the Halliburton divisions
in London and in Houston. Indeed, they now plan to
design a South American project in Brazil and to designreview it from Leatherhead.
These groups find true scale for structures is important for detailed understanding. This is, I believe, the
major advantage that HMDs, caves, and panoramic displays have over so-called desktop VR. I think the advantage justifies the extra cost when dealing with complex
structures.
Second, system latency—a major concern in the VR
laboratory—is not necessarily a showstopper for this
application, because viewpoint motion is slow and infrequent, due to the intense study going on.
VR today remains an expensive technology, more
because of the people, the models, and the software
than because of the image generation, tracking, and display equipment. Therefore, adoption of the technology
for design tasks happened first in applications where the
value of design is greatest: mass-produced vehicles and
exceedingly complex structures such as submarines and
process plants.
The Electric Boat Division of General Dynamics Corporation, which designs and builds nuclear submarines,
uses VR principally for multidisciplinary design review.
The configuration consists of a large conference room
with a panoramic monocular projection system. A single user specifies travel using a mouse, and an SGI Onyx
Infinite Reality Engine generates imagery. Electric Boat
makes heavy use of four such high-security VR rooms at
its Groton, Connecticut, development laboratories. My
hosts there were Don Slawski and Jim Boudreaux.
Submarine design is done in and controlled with
Catia. Approximately two-thirds of the total design
effort (for the vessel, not including its nuclear power
plant) goes into the design of piping, wiring, and ductwork. The visualizable model of a submarine contains
millions of polygons. This model is derived from the
Catia model, transmitted to the visualization file system
by “sneakernet,” and displayed using Deneb’s visualization software.
A typical design review session includes not only the
various engineering groups involved in a particular
design area, but also manufacturing, capital tooling,
maintenance, and operations people. During a session
the group will usually move to a particular local area
and spend many minutes studying it. Viewpoint changes
may happen every few minutes to aid study, but not continually. Representatives of each discipline evaluate the
proposed design from their own particular points of
view, discuss problems, and propose fixes. After the
meeting, the responsible engineer details the fixes and
enters the changes into the Catia system manually,
through the normal change-control process. The normal structural, stress, acoustic, and other analyses are
run, usually as batch operations, on the changed design.
In particular, the oft-imagined scenario in which the
visiting admiral says “Move that bulkhead one foot aft!”
and the change is immediately effected in the master
model is not how things are done, of course. Such a
thing would be technically very difficult for the VR system, to be sure. But such a scenario would not happen
even with magical technology, because design discipline
demands the human and computational analysis of all
the interactions and effects of any proposed change.
Design review at John Deere
The experience of the Construction Machinery Division of John Deere as it has moved toward virtual prototyping proves instructive. Pilot studies of VR at the
Moline, Illinois division began in 1994, but nothing
much really happened until a key technical person
joined in 1996. The configuration is rather typical: an
SGI Onyx Infinite Reality, a head-mounted display, an
Ascension Flock of Birds tracking system, Division’s
dVise software, and the Jack human model for
ergonomic studies. Zones of reach and zones of comfort, provided by Engineering Animation Inc.’s Pro-Engineering software, are shown superimposed on the
virtual models.
Jerry Duncan and Mac Klinger served as my infor-
IEEE Computer Graphics and Applications
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
23
Special Report
over-tessellate curved parts, they
wish for a suite of tools for polyhedral simplification and for file management for large models and many
versions of models.
Lessons. The foresight and
activity of an advocate at the corporate technical level facilitated experimentation and then adoption of VR
technology at the division level. The
vision at the corporate level for John
Deere as a whole encompasses not
only VR for rapid virtual prototyping of products, but for analysis and
evaluation as well as design. And the
vision encompasses not only products, but also manufacturing
processes, for more Deere people
are engaged in the manufacturing
process than in product design.
An important insight is that the
farther downstream a design gets from its conceiver, the
better the visuals must be to convey an accurate, internalized perception of the design to those who must
build, maintain, and use it. The creating engineer can
visualize from a sketch, and the experienced toolmaker
from an engineering drawing, but a factory-floor foreman or an operations person may be unable to visualize from anything short of a detailed, realistic picture.
As all the examples above indicate, each user of VR
for design has found that it radically facilitates communication among the design team. This may be the most
dramatic and cost-effective result.
John Deere has for some time collaborated with the VR
research group at Iowa State University, where Carolina
Cruz-Neira showed me around. The company sees the
Iowa State work as a technical forerunner and groundbreaker for Deere’s own in-house applications. The Iowa
State group, for example, has a Deere tractor buck mounted in a cave configuration, with a detailed model of a particular Iowa farm, acquired by GPS surveys. John Deere
currently operates a more modest configuration for its
production work, with exploratory work done at the Iowa
State facility. I saw this pattern emerging in several places,
including automobile companies that use the Fraunhofer
Institutes in Stuttgart and Darmstadt as the sites of their
explorations. It makes a lot of sense to me. University
research groups should be in the trailblazing business,
and the degree to which others then follow those trails
indicates the excellence and relevance of the research.
Courtesy of NASA
3 NASA-Houston’s “Charlotte” Virtual
Weightless Mass
lets astronauts
practice handling weightless
massy objects.
mants. Four wins to date reflect the sort of savings resulting from the nontrivial investment in VR. The first study
was a styling study for a new machine, normally done
with a sequence of fiberboard mockups. The VR model
displaced two-thirds, but not all, of the usual mockups
and saved $80K in mockup cost—real savings only if you
already have a CAD model of the prototyped object.
The second win was a safety technology evaluation
formerly done on an iron prototype, successfully performed in VR. This evaluation studies, among other
things, encumbrances and egress in emergencies. A sixperson jury performs the evaluation, using a formal protocol. It normally takes two days, because each jury
member climbs into the prototype, evaluates the next
group of items on the checklist, and climbs back out.
The VR evaluation had the whole jury viewing the video
projection on a screen while one person, wearing the
head-mounted display, did the various maneuvers.
Obviating the serialized access to the cab shortened the
whole process to three hours. The whole jury saw and
discussed each aspect from a common eyepoint, focusing the discussion. The savings were both man-hours
and the construction of the iron prototype.
The third win was exactly the kind hoped for with virtual prototype design review. The issue was steps and
handholds, for all kinds of body sizes—a matter that
seems minor but makes big differences in usability.
Three entire rounds of review and change were effected in five days—with no iron-cutting.
Finally, Deere was able to conduct early customer tests
of the new 450-H Crawler-dozer using the virtual prototype, well before they could have done them with an
iron prototype.
Wish list. The most strongly desired tools are geometry manipulation tools, ways of easily specifying interactions with the design. This includes collision detection
and collision response. The great desire is for interfaces
simple enough for the occasional user to participate in
model changing. Then, since they too find that CAD tools
24
Astronaut training at NASA
Training applications in routine production status
proved hard to find, although quite a few are far along
in prototype status. The training work with the longest
experience—several years as a production operation—
seems to be that of NASA-Houston for training astronauts for extra-vehicular activity, where Bowen Loftin
and David Homan served as my hosts.
It’s hardly surprising that NASA’s astronaut training
should lead the way; the training has very high value,
November/December 1999
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
Courtesy of Georgia Institute of Technology
4 Vietnam war
simulation at
the Atlanta
Veterans
Administration
Hospital (a) in
use. (b) Imagery
seen by the
user.
and the alternatives to VR technology are few and poor.
Of course, much training can use vehicle mockups.
Weightless experience can be gained in swimming pools
and 30-second-long weightless arcs in airplanes.
Nonetheless, extra-vehicular activity is very difficult to
simulate.
A difficult skill, for which VR training has proven very
powerful, is flying about in space using the back-mounted flight unit. It is like no earthly experience. Newton’s
laws apply ideally—with no drag, an object in motion
or rotation continues forever unless stopped. The flight
unit is designed principally as an emergency device for
use if an astronaut’s tether breaks; velocities are very
slow. The NASA VR configuration for this is rather standard: an SGI Onyx Infinite Reality for imagery generation, head-mounted displays, and magnetic tracking.
An astronaut said the resulting system was the most
faithful flight simulator he had used.
Moving around on the outside of a space vehicle is
another unearthly skill. The VR system lets astronauts
practice the careful planting of hands and feet in rockclimbing fashion. Difficult and unprecedented team
tasks, such as correcting the Hubble telescope mirror’s
optics, made new training demands. The additional
unearthly experience for such tasks is the team-coordinated moving of massy but weightless objects. The
dynamics are, of course, totally unfamiliar, and viscous
damping seriously confounds underwater simulation.
For this training, NASA augments visual simulation with
a unique haptic simulator called “Charlotte” after the
spider of the same name. A real but very light 2-foot
cubical box attaches to motors on the corners of an 8foot cubical frame, as shown in Figure 3. As pairs of
astronauts move the object by its handles, the system
simulates the dynamics and drives the motors appropriately. Users report very high fidelity for masses of 300
pounds and up.
Lessons. Interaction with objects in virtual worlds
still challenges VR researchers. So much of our interaction with objects depends on how they feel—purely visual simulation misses realism much too far. This is critical
for some training tasks; we really do not know how
important it is for design tasks.
Early adoption of VR, even with less-than-satisfactory technologies, enabled NASA to get the years of experience that brought their applications to their present
effectiveness.
Collaboration with researchers at the University of
Houston has worked much like the John Deere-Iowa
State collaboration. The university can do wider exploration and trailblazing; the mission agency does the
focused work that leads to production applications.
Psychiatric treatment at Georgia Tech-Emory
Research collaborators at Georgia Institute of Technology and the Emory University Medical School have
explored the use of VR for psychiatric treatment. Larry
Hodges, Barbara Rothbaum, and David Ready hosted
my visit there.
The success of this research has led to the development and fielding of a COTS system by Virtually Better,
Inc. They offer a system for about $18K, approximately
$8K for the hardware and $10K for software for one
application. Eight installations are already in production around the US and one in Argentina, all routinely
used by practicing psychiatrists and psychologists. These
practitioners do not have computer support locally; any
necessary support comes by telephone from Atlanta.
The hardware configuration consists of a PC, a graphics accelerator card, a Virtual Research V-6 or V-8 headmounted display, and a magnetically tracked bat. A good
deal of attention has gone to audio quality; the visuals
look rather cartoonish.
The most popular and cost-effective application treats
fear-of-flying. The conventional treatment has the practitioner and patient together make multiple trips to an
airport, sit on an airplane mockup, sit on an airplane,
fly a short hop, and so on. Much expensive practitioner
time disappears in travel. The effectiveness of the VR
treatment seems just as good as the conventional one,
although there are no formal studies yet. The cost is radically lower. Virtually Better also offers fear-of-heights
and fear-of-public speaking simulation packages.
The most dramatic procedure offered, in routine use
at the Atlanta Veterans Administration Hospital, treats
post-traumatic stress disorder for Vietnam War veterans. Physiological monitoring of the patient augments
IEEE Computer Graphics and Applications
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
25
Special Report
Courtesy of Russell Taylor
5 View of UNC
nanoManipulator connected
to an atomic
force microscope.
the VR configuration, to give an independent measure
of his emotional stress level. The psychologist gently
leads the patient into a simulated Vietnam battle scene
(see Figure 4), step-by-step recreating the situation
where the patient “locks up” in reliving his stress experience. By leading the patient completely through the
scene and out the other side, the psychologist aims to
help the patient learn how to get himself out of the damaging patterns.
So far, the treatment seems to help the patients who
persevere. About half of the first 13 opted out, perhaps
because of the realism of the recreated experiences.
Studying tapes of typical sessions, I was struck by how
completely the patients are present and involved in the
simulated Vietnam situation. It hurts to watch them
relive painful experiences.
A second installation of this application is under way
at the VA hospital in Boston.
Lessons. The big lesson for me was the power of aural
VR for reproducing an overall environment. Larry
Hodges, the computer scientist on the team, thinks that
audio quality is, in several of their applications and experiments, consistently more important than visual quality.
His Vietnam simulation certainly supports that opinion.
Indeed, the Fraunhofer IAO (Industrial Engineering
Institute) at Stuttgart has a prototype VR application,
the routine quality testing of electric drills after manufacture, in which the only VR environment is aural. A
sound shower displays the sound field of a well-made
drill all over the test cell, facilitating aural comparison.
It uses no visuals at all.
I am also impressed with the importance of full-scale
immersion for the Vietnam application. It is hard to
believe that desktop VR would even begin to achieve the
same effects.
nanoManipulator at UNC-Chapel Hill
Probe microscopes, including atomic force microscopes, scanning tunneling microscopes, and near-field
26
optical microscopes, form images by scanning a probe
across the surface of a sample, yielding one or more
scalar functions of the two dimensions. Resolution to
the nanometer can be obtained; even individual atoms
can be imaged under best conditions. As significantly,
the tip of the probe can modify the sample, by mechanical or electrical action.
Researchers at the University of North Carolina at
Chapel Hill applied VR technology to the task of making the using scientist think and act as if shrunk by a factor of 105 to 106 and present on the surface, in a VR
system called the nanoManipulator. (An alternative way
of thinking about it is that we expand the sample by such
a factor, to laptop sizes.) The scientist controls the viewpoint and the lighting on the dynamically updated
image of the sample, as the microscope continues to
scan. He may, if he chooses, suspend scanning, position
the probe to a particular spot, and make measurements
there. Alternatively, he may take control of the probe
and effect the modifications as if he were scratching the
surface directly. Russell Taylor and Richard Superfine
are leading this project.
Perception is effected two ways: through a stereo visual image rendered on an SGI Onyx Infinite Reality
engine and displayed to a head-tracked observer on a
workbench or a desktop monitor, and through a Sensable Systems Phantom haptic display (see Figure 5). The
user holds this motor-controlled stylus like a pen; sensors yield 4D measurements of position and pose of the
tip. Motors provide three to six degrees of force output.
The nanoManipulator currently presents three degrees
of force output. Moving the Phantom manually controls
tip position and action.
The haptic display and control prove essential for
manipulations. Since the probe can either image or
scrape, but not both at the same time, the scientist is
blind while manipulating the surface with the microscope probe. The standard technique has the scientist
image the sample, manipulate the probe while working blind but seeing the previous image, then image
again to find out what really happened. Although the
probe cannot produce images while being used for
manipulations, it does produce the control signals that
display as vertical or lateral forces. So the scientist can
feel what he’s doing even when he cannot see what’s
going on.
The sensation of interacting with a real macroscopic
sample on a workbench is very strong. For this application, the size and small field of view of a desktop monitor does not hinder perception because the virtual object
is created “full size,” as if on a workbench
Replications of the nanoManipulator system have
been installed in four locations in the US and Europe.
In routine daily use, they produce science published by
physicists, chemists, gene therapists, and others.7
Lessons. This application illustrates and emphasizes
the fruitfulness of haptics in a VR configuration. It offers
an almost ideal application of the Phantom, which can
display only the forces on a point probe. Since the microscope can only measure forces on a point probe, the two
match well.
November/December 1999
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
Therefore, the methods of interaction seem quite natural for both probe motions and for head, sample, and
light motions to improve viewing. Early research work
showed that realistic haptic rendering requires update
rates greater than 500 updates per second. The Phantom runs at 1,000 updates per second, which seems
quite satisfactory.
Second, each VR display mode has some application
for which it provides the optimal solution. This one is a
natural for a projection-display workbench, a workbench equipped with a haptic display.
Hot open challenges
Although VR has crossed the high pass from “almost
works” to “barely works,” many challenges remain both
in the enabling technologies and in the systems engineering and human factors disciplines. The tasks that
seem most crucial to me follow:
Technological:
■ Getting latency down to acceptable levels.
■ Rendering massive models (> 1 M polygons) in real
time.
■ Choosing which display best fits each application:
HMD, cave, bench, or panorama.
■ Producing satisfactory haptic augmentation for VR
illusions.
Systems:
■ Interacting most effectively with virtual worlds:
■ Manipulation
■ Specifying travel
■ Wayfinding
■ Making model worlds efficiently:
■ Modeling the existing world—image-based
techniques look promising
■ Modeling nonexisting worlds—through
CAD byproducts or hard work
■ Measuring the illusion of presence and its operational
effectiveness
How the field addresses these challenges will dramatically affect the continuing success and speed of
adoption of VR. I look forward to seeing substantially
more projects move from prototype to production status in the near future.
■
Acknowledgments
A major part of this report was given as a keynote
address at the Virtual Reality 99 conference.
My greatest thanks go to the correspondents who sent
me descriptions of their VR projects, and most especially
to those who welcomed me to their installations and
shared with me their experiences and the lessons they
drew from them. The project descriptions list the principal hosts at each installation; each had colleagues who
also shared their knowledge and time. I am most grateful to each of them.
An imaginative grant from the Office of Naval
Research, arranged by Larry Rosenblum, made possible the travel by which I gathered the solid first-hand
knowledge of applications that are really in production
status. I also thank my faculty, staff, and student colleagues at Chapel Hill, who have taught me many
lessons about virtual reality and who continually amaze
me with their creativity and accomplishment. Henry
Fuchs, in particular, has shared the vision and inspired
my efforts by his own immense imagination.
References
1. I.E. Sutherland, “The Ultimate Display,” invited lecture,
IFIP Congress 65. An abstract appears in Information Processing 1965: Proc. IFIP Congress 65, Vol. 2, W.A. Kalenich,
ed., Spartan Books, Washington, D.C., and Macmillan,
New York, pp. 506-508. A similar, but quite complete,
paper is Sutherland’s “Computer Displays,” Scientific American, Vol. 222, No. 6, June, 1970, pp. 57-81.
2. F.P. Brooks, “Is There Any Real Virtue in Virtual Reality?,”
public lecture cosponsored by the Royal Academy of Engineering and the British Computer Society, London, 30 Nov.
1994; http://www.cs.unc.edu/~brooks.
3. S. Coleridge, Biographia Literaria, Ch. 1, in The Collected
Works of Samuel Taylor Coleridge, Vol. 7, J. Engell and W.J.
Bate, eds., Routledge & Kegan Paul, London, and Princeton
University Press, Princeton, 1983.
4. N.I. Durlach and A.S. Mavor, eds., Virtual Reality—Scientific and Technological Challenges, National Research Council, Washington, D.C., 1995.
5. G. Burdea and P. Coiffet, Virtual Reality Technology, John
Wiley, New York, 1994.
6. R.L. Holloway, Registration Errors in Augmented Reality Systems, PhD dissertation, Department of Computer Science,
University of North Carolina at Chapel Hill, 1995.
7. M.R. Falvo et al., “Bending and Buckling of Carbon Nanotubes under Large Strain,” Nature, Vol. 389, Oct. 1997,
pp. 582-584.
Frederick P. Brooks, Jr. is
Kenan Professor of Computer Science
at the University of North Carolina
at Chapel Hill, where he founded the
department in 1964. His current
research field is interactive 3D computer graphics. He received his PhD
in 1956 in what would today be called computer science,
under Howard Aiken at Harvard. His AB was in physics
from Duke.
While Brooks worked for IBM, he was finally project
manager for the development of the System/360 family of
computers and of its software, Operating System/360, for
which he was awarded the National Medal of Technology
and the IEEE John von Neumann Medal. His recent books
are The Mythical Man-Month: Essays on Software Engineering (20th Anniversary Edition, Addison-Wesley,
Reading, Mass., 1995) and, with G.A. Blaauw, Computer Architecture: Concepts and Evolution (Addison-Wesley, 1997). He is a fellow of the IEEE and a member of the
National Academy of Engineering.
Readers may contact him at brooks@cs.unc.edu.
IEEE Computer Graphics and Applications
Authorized licensed use limited to: IEEE Xplore. Downloaded on December 11, 2008 at 08:06 from IEEE Xplore. Restrictions apply.
27